The Clinical Outcome of Different Total Knee Arthroplasty Designs in One-Stage Revision for Periprosthetic Infection

Background: Despite the growing number of studies reporting on the best surgical treatment in the management of periprosthetic joint infection, there are no robust data regarding the type of infected prosthesis before any kind of exchange arthroplasty. To overcome these shortcomings, we asked the following questions: (1) What is the survivorship of nonhinged and hinged knee implants after one-stage exchange arthroplasty and (2) what is the functional outcome after one-stage exchange procedure focusing on knee prostheses and the type of prior infected knee implant. In a secondary radiographic analysis, we also investigated if (3) the type of femoral bone morphology measured by the inner femoral diameter influences the rate of aseptic failures also in patients with periprosthetic joint infection. Methods: Between January 2011 and December 2017, we performed a retrospective designed study including 211 patients with infected knee prostheses. After all, seventy-six percent (161 of 211 patients) were available for final data analysis. These patients were divided into four groups as per the performed implant revision: (1) bicondylar total knee arthroplasty to rotating hinge implant, (2) rotating hinge to rotating hinge implant, (3) rotating hinge to full hinge implant, and (4) full hinge to full hinge implant. The mean follow-up (FU) was six years (range 3 to 9; standard deviation = 1.9), whereas a minimum FU of three years was required for inclusion. Survivorship and group analysis were performed, and the functional outcome was assessed using postoperative Oxford Knee Scores at the latest FU (60-point scale with lower scores representing less pain and greater function). Furthermore, in all cases, femoral bone morphology was determined as per the Citak classification system. Results: At the final FU, the overall surgical revision rate was 23% (37/161 patients) with nine percent (15/ 161 patients) suffering a periprosthetic joint infection relapse. Group 1 consisted of 51, group 2 consisted of 67, group 3 consisted of 24, and group 4 consisted of 19 patients. The lowest overall revision rate was found in group 2 (16%, n = 11), compared with 28% (n = 14) in group 1, 29% (n = 7) in group 3, and 26% (n = 5) in group 4; however, no significant differences were found (P = .902). The functional outcome (Oxford Knee Score) was clinically constant in all groups, with 32 points in group 1, 37 points in group 2, 33 points in group 3, and 35 points in group 4 (P = .107). Concerning the number of patients with aseptic loosening as per bone morphology, 74% (14/19) of all aseptic loosening cases appeared in femoral bone type C morphologies according to Citak (75% in group 1, 56% in group 2, 100% in group 3, and 100% in group 4). Conclusions: The results obtained suggest a generally high overall revision rate (25%) with a good infection control rate (91%). Although we were unable to work out a specific group of patients with a statistically significant differing outcome, it is interesting to see that hinged implants can reach more or less the same functional outcome and revision rates as nonhinged implants, when it comes to revision surgeries. In this study, a relatively high number of aseptic failures contributed to a high overall revision rate. In this context, the bone morphology, measured as per the Citak classification system, could be confirmed as a risk factor for aseptic failures also in septic patients. Therefore, further research might focus on revision knee implant design.